Superwhisper alternatives: 8 free & open-source picks
Curated, verified, and actually free. No trials, no crippled tiers. Star counts and last-activity dates checked by hand, not scraped and forgotten.
Voice cloning studio attached to a global dictation hotkey; overkill, free overkill.
licenseMIT
runson your machine
installPrebuilt macOS or Windows app, Docker, or a source build on Linux
enginesWhisper and Whisper Turbo speech recognition through MLX or PyTorch, local Qwen3 text cleanup, and multiple local text-to-speech and voice-cloning engines.
dataProjects, transcripts, model metadata, and settings live in a local SQLite database with local audio and voice assets.
the catch vs SuperwhisperVoicebox is a heavyweight voice studio whose dictation hotkey is only one module, bringing larger models and more complexity than Superwhisper's focused utility.
setupThe first run is a large model-and-compute setup rather than a tiny dictation utility; Linux users currently need a source build.
facts verified 2026-08-10
Press a key, talk and get plain text in the app you were already using. No cleanup magic, no invoice.
licenseMIT
runson your machine
installDMG, MSI, deb, rpm, or AppImage
enginesLocal Whisper Small, Medium, Turbo, and Large models; NVIDIA Parakeet v3; Silero VAD; optional cleanup through local or configured LLM and OpenRouter-compatible providers.
dataHistory is stored in a local SQLite history.db database and recordings as local WAV files; configured cleanup providers receive the transcript sent to them.
the catch vs SuperwhisperHandy is transcription-first and lacks Superwhisper's mature mode system and consistently polished context-aware cleanup.
setupThe first run needs microphone and text-injection permissions plus a model download; on Linux, automatic insertion may require xdotool, wtype, or dotool depending on the desktop.
facts verified 2026-08-10
Fast Mac dictation with local cleanup; even the fancy part stays on the machine.
licenseGPL-3.0
runson your machine
installSigned macOS download or Homebrew cask
enginesLocal engines include NVIDIA Nemotron Speech, Parakeet Flash/TDT, Cohere Transcribe models, Apple Speech, and Whisper; optional cleanup can use the local Fluid Intelligence model or OpenAI, Groq, and custom compatible endpoints.
dataTranscripts, optional history, recordings, dictionaries, and settings stay on the Mac; API keys use Keychain, and optional cloud cleanup sends text to the selected provider. Exports can be created locally.
the catch vs SuperwhisperFluidVoice is macOS-only and newer, with fewer proven ready-made dictation modes and less cross-device polish than Superwhisper.
setupIt requires macOS 15, microphone and Accessibility permissions, and a model download; Intel Macs are limited to the Whisper path.
facts verified 2026-08-10
A cross-platform hotkey-to-cursor app with local models, cleanup, history and meeting capture.
licenseMIT
runsyour machine or their cloud
installNative macOS, Windows, or Linux installer; optional account enables cloud services
enginesLocal Whisper, NVIDIA Parakeet, and llama.cpp-based processing; optional managed or bring-your-own-key cloud paths use providers such as OpenAI and Groq.
dataLocal mode stores transcriptions, notes, agent data, and preferences in a local SQLite database, with optional local recordings; cloud sync is optional, and cloud transcription audio is described as transient rather than retained.
the catch vs SuperwhisperThe free local path is less turnkey, while cloud speed, synchronization, and several advanced AI conveniences move into paid or bring-your-own-provider territory.
setupThe free local path needs permissions and a model download; cloud-speed, synchronization, and some advanced AI conveniences require an account, provider key, or paid plan.
facts verified 2026-08-10
System-wide dictation, local Whisper, cleanup and a glossary; cloud convenience is the paid part.
licenseAGPL-3.0
runsyour machine or their cloud
installNative macOS, Windows, or Linux installer; on-premises deployment is also offered
enginesLocal Whisper; cloud or BYOK connections include Groq, OpenAI, Anthropic, Google, DeepSeek, OpenRouter, Azure, and compatible endpoints, with local or provider-backed cleanup.
dataLocal history, audio, settings, and encrypted keys live in a local SQLite database; optional cloud synchronization stores account, dictionary, and tone data in Firebase, while cloud audio is described as transient.
the catch vs SuperwhisperIts free local or BYOK route is less zero-configuration, and the smooth cloud synchronization and fast managed transcription path are paid.
setupLocal use requires permissions and a model download; the smoother cloud transcription and sync path needs an account, provider credentials, or the paid service.
facts verified 2026-08-10
Hold a key, talk and get cleaned text at the cursor; the local meeting recorder comes along for free.
licenseMIT
runson your machine
installSigned macOS DMG or Homebrew cask
enginesLocal speech models include Parakeet variants, Cohere Transcribe, Nemotron, SenseVoice, Qwen3 ASR, Indic ASR, Gemma, and Whisper; cleanup and summaries can use OpenAI, OpenRouter, ChatGPT OAuth, Ollama, LM Studio, or custom endpoints.
dataThe app stores its library in a local SQLite database under the macOS Application Support folder and recordings on disk; optional iCloud sync is text-only and recordings do not sync. Exports include PDF, Markdown, and JSON.
the catch vs SuperwhisperMuesli is Apple-silicon Mac-only and exposes more model and provider plumbing instead of Superwhisper's more polished, ready-made dictation experience.
setupIt requires Apple-silicon macOS 14.2 or later, microphone and Accessibility permissions, and at least one local model download or external provider configuration.
facts verified 2026-08-10
Control, talk, Control, paste. It has no settings because there is nothing to upsell.
licenseGPL-3.0
runson your machine
installDownload the signed and notarized macOS app, then install its separate local speech-engine dependencies
enginesThe local coli speech-recognition engine; the current public setup does not identify a supported cloud backend.
dataIt keeps no transcript history: captured audio is processed locally, the result is placed in the clipboard or active app, and downloaded model files live under ~/.coli/models.
the catch vs SuperwhisperTypeNo deliberately omits cleanup, modes, history, dictionaries, and settings, so it replaces raw transcription rather than Superwhisper's polished rewriting workflow.
setupDespite the polished app bundle, setup requires Node.js LTS, ffmpeg, a global npm install of @marswave/coli, and a separate model download.
facts verified 2026-08-10
Unlimited local dictation behind a polished installer; cloud convenience is the bit it charges for.
licenseproprietary, free
runsyour machine or their cloud
installNative macOS, Windows, Linux, or iPhone app
enginesFree local transcription uses Whisper and Parakeet; bring-your-own-key cloud options include OpenAI, Deepgram, and Groq, while Spokenly's managed cloud and AI cleanup are paid.
dataOffline mode keeps audio and transcripts on the device; a selected cloud provider receives audio for cloud transcription and the provider's retention rules apply. Spokenly does not publish the local database or file format.
the catch vs SuperwhisperUnlimited local transcription is free, but the managed cloud speed, AI cleanup, and synchronization that make the experience feel effortless are paid.
setupLocal use needs microphone and Accessibility permissions plus a model download; BYOK cloud use requires creating and protecting a provider key.
facts verified 2026-08-10
last updated 2026-08-10 · no votes, no pay-to-list · just what's real
VoiceInkThe code is open, but the usable download is a trial; the free path is compiling it yourself in Xcode.
| tool | license | running it | platforms | stars | active |
|---|---|---|---|---|---|
| Voicebox | MIT | one-click install | macos, windows | 49,500 | 2026-06 |
| Handy | MIT | one-click install | macos, windows, linux | 28,900 | 2026-08 |
| FluidVoice | GPL-3.0 | one-click install | macos | 9,400 | 2026-07 |
| OpenWhispr | MIT | one-click install | macos, windows, linux | 5,223 | 2026-08 |
| Voquill | AGPL-3.0 | one-click install | macos, windows, linux, ios | 997 | 2026-08 |
| Muesli | MIT | one-click install | macos | 890 | 2026-08 |
| TypeNo | GPL-3.0 | one-click install | macos | 883 | 2026-05 |
| Spokenly | proprietary, free | one-click install | macos, windows, linux, ios | — | — |
They pay for the low-friction menu-bar experience and dictation profiles. If none of that applies to you, any pick above saves $101.88 a year.
Is there a free alternative to Superwhisper?
Yes: Voicebox, Handy, FluidVoice and 5 more. Every tool on this page is either open source or genuinely free to use long-term: no trials, no crippled tiers.
Should I just build my own Superwhisper?
Our verdict is SÍ. A push-to-talk recorder that transcribes and pastes text into the active app is very buildable for one person, especially on macOS. If you'd rather build, the exact one-shot prompt is free at puedovibecodearlo.app/superwhisper.
